Path: utzoo!attcan!uunet!mcvax!ruuinf!piet
From: piet@ruuinf.UUCP (Piet van Oostrum)
Newsgroups: comp.lang.misc
Subject: Re: Algol 60 vs Algol 68 (was "stack machines (Burroughs)")
Message-ID: <492@ruuinf.UUCP>
Date: 16 Jun 88 15:46:23 GMT
References: <1521@pt.cs.cmu.edu> <1532@pt.cs.cmu.edu> <476@pcrat.UUCP> <2868@louie.udel.EDU> <370@dlscg1.UUCP> <3147@polyslo.UUCP> <10064@tekecs.TEK.COM> <1988Jun11.200757.12285@light.uucp>
Followup-To: comp.lang.misc
Organization: Univ of Utrecht, Dept of CS
Lines: 48
In-reply-to: bvs@light.uucp's message of 12 Jun 88 03:07:54 GMT

In article <1988Jun11.200757.12285@light.uucp> bvs@light.uucp (Bakul Shah) writes:

   In article <10064@tekecs.TEK.COM> Andrew Klossner writes:
   > ...
   >There really isn't any "beyond" to Algol 68 since the 1975 Revised
   >Report.  It's a dead language.  And that's too bad; while its model of
   >computation is distant from that of real machines (making it an
   >inappropriate language for most low level systems programming), it does
   >an admirable job in its stated domain of algorithmic description, and
   >is great for applications programming.

   Actually Algol 68's model quite closely matches real machines and
   the language can be compiled to generate code as efficient as
   compiled C; though I doubt there are any modern, optimizing
   compilers for it.  The language is verbose and the its syntax
   rather od (!) but it can be used quite effectively for systems
   programming.  I recall the CAP operating system (for the Cambridge
   CAP computer) was written in Algol 68.

   I wonder how much of C design was influenced by Algol 68 as most
   of K&R C seems to map almost directly to Algol 68 and where C
   differs is usually where the compiler's job is simplified.  Come
   to think of it, a major subset of Algol 68 with a new and concise
   syntax (sort of like C's) can make a very elegant, type safe and
   well rounded language.

I wrote major parts of both an Algol 68 and an Algol 60 compiler, so I
assume I know a bit of the difficulties. Regarding ALGOL 68 and C, I think
there are a lot of ideas from Algol 68 in C, but -- as in Pascal -- C left
out a lot of the hard things. The syntax of C is much closer to Algol 68
than Pascal. Also some Algol 60 syntax crept into C, e.g. function headers.

Now any of the left out features from Algol 68 (e.g. heap variables,
dynamic arrays) are not very difficult to add to implement, BUT if you add
two or more features at the same time they are going to interact in a
terrible manner. The complexity of the resulting compiler is almost
exponential to the number of added features. One example: dynamic arrays
are easy: Algol 60 had it, see also the discussion on alloca in this
newsgroup. Unions are easy: C has them. The combination of dynamic arrays
and union however forces you to use the heap, and hence garbage collection.

Although I must say I like garbage collection very much, it assumes that
you have a perfectly secure language, which C isn't.

-- 
Piet van Oostrum, Dept of Computer Science, University of Utrecht
Padualaan 14, P.O. Box 80.089, 3508 TB Utrecht, The Netherlands
Telephone: +31-30-531806              UUCP: ...!mcvax!ruuinf!piet